Spherical Headgear HRIR Compilation of the Neumann KU100 and the Head acoustics HMS II.3

[1] C. Pörschmann, J. M. Arend, and R. Gillioz, “How wearing headgear affects measured head-related transfer functions,” in Proceedings of the EAA Spatial Audio Signal Processing Symposium, 2019, pp. 49–54. DOI link: https://doi.org/10.25836/sasp.2019.27 Files also available at in SOFA file format at sofacoustics.org/data/database/thk/ _______________________________________________________________________________________________________ The spatial representation of sound sources is an essential element of virtual acoustic environments (VAEs). When determining the sound incidence direction, the human auditory system evaluates monaural and binaural cues, which are caused by the shape of the pinna and the head. While spectral information is the most important cue for elevation of a sound source, we use differences between the signals reaching the left and the right ear for lateral localization. These binaural differences manifest in interaural time differences (ITDs) and interaural level differences (ILDs). In many headphone-based VAEs, head-related transfer functions (HRTFs) are used to describe the sound incidence from a source to the left and right ear, thus integrating both monaural and the binaural cues. Specific aspects, like for example the individual shape of the head and the outer ears (e.g. Bomhardt, 2017), of the torso (Brinkmann et al., 2015), and probably even of headgear (Wersenyi, 2005; Wersenyi, 2017) influence the HRTFs and thus probably as well localization and other perceptual attributes. Generally speaking, spatial cues are modified by headgear, for example by wearing a baseball cap, a bicycle helmet, or a head-mounted display, which nowadays is often used in VR applications. In many real life situations, however, a good localization performance is important when wearing such items, e.g. in order to determine approaching vehicles when cycling. Furthermore, when performing psychoacoustic experiments in mixed-reality applications using head-mounted displays, the influence of the head-mounted display on the HRTFs must be considered. Effects of an HTC Vive head-mounted display on localization performance have already been shown in Ahrens et al. (2018). To analyze the influence of headgear for varying directions of incidence, measurements of HRTFs on a dense spherical sampling grid are required. However, HRTF measurements of a dummy head with various headgear are still rare, and to our knowledge only one dataset measured for an HTC Vice on a sparse grid with 64 positions is freely accessible (Ahrens, 2018). This work presents high-density measurement data of HRTFs from a Neumann KU100 and a HEAD acoustics HMS II.3 dummy head, either equipped with a bicycle helmet, a baseball cap, an Oculus Rift head-mounted display, or a set of extra-aural AKG K1000 headphones. For the measurements, we used the VariSphear measurement system (Bernschütz, 2010), allowing precise positioning of the dummy head at the spatial sampling positions. The various HRTF sets were captured on a full spherical Lebedev grid with 2702 points. In our study, we analyze the measured datasets in terms of their spectrum, their binaural cues, and regarding their localization performance based on localization models, and compare the results to reference measurements of the dummy heads without headgear. The results show that differences to the reference without headgear vary significantly depending on the type of the headgear. Regarding the ITDs and ILDs, the analysis reveals the highest influences for the AKG K1000. While for the Oculus Rift head-mounted display, the ITDs and ILDs are mainly affected for frontal directions, only a very weak influence of the bicycle helmet and the baseball cap on ITDs and ILDs was observed. For the spectral differences to the reference the results show maximal deviations for the AKG K1000, the lowest for the Oculus Rift and the baseball cap. Furthermore, we analyzed for which incidence directions the spectrum is influenced most by the headgears. For the Oculus Rift and the baseball cap, the strongest deviations were found for contralateral sound incidence. For the bicycle helmet, the directions mostly affected are as well contralateral, but shifted upwards in elevation. Finally, the AKG K1000 headphones generally has the highest influence on the measured HRTFs, which becomes maximal for sound incidence from behind. The results of this study are relevant for applications where headgears are worn and localization or other aspects of spatial hearing are considered. This could be the case, for example in mixed-reality applications where natural sound sources are presented while the listener is wearing a head-mounted display, or when investigating localization performance in certain situations, e.g. in sports activities where headgears are used. However, it is an important intention of this study to provide a freely available database of HRTF sets which is well suited for auralization purposes and which allows to further investigate the influence of headgear on auditory perception. The HRTF sets will be publicly available in the SOFA format under a Creative Commons CC BY-SA 4.0 license. ________________________________________________________________________________________________________ Contact: Christoph Pörschmann TH Köln - University of Applied Sciences Institute of Communications Engineering Department of Acoustics and Audio Signal Processing Betzdorfer Str. 2, D-50679 Cologne, Germany https://www.th-koeln.de/akustik _________________________________________________________________[1] C. Pörschmann、J. M. Arend 与 R. Gillioz，《佩戴头部配饰对实测头相关传输函数（Head-related Transfer Functions, HRTFs）的影响》，收录于2019年EAA空间音频信号处理研讨会论文集，第49-54页。DOI链接：https://doi.org/10.25836/sasp.2019.27，数据亦可通过SOFA格式（SOFA）于sofacoustics.org/data/database/thk/获取。 _______________________________________________________________________________________________________ 声源的空间表征是虚拟声环境（Virtual Acoustic Environments, VAEs）的核心要素。人类听觉系统在判定声音入射方向时，会评估由耳廓与头部外形引发的单耳线索与双耳线索。其中，频谱信息是判定声源仰角的核心线索，而侧向定位则依赖于左右耳接收信号间的差异。这类双耳差异体现为耳间时间差（Interaural Time Differences, ITDs）与耳间声级差（Interaural Level Differences, ILDs）。 在诸多基于耳机的虚拟声环境中，研究人员常采用头相关传输函数（Head-related Transfer Functions, HRTFs）来描述声源至左右耳的声音传输过程，以此整合单耳与双耳线索。头部与外耳的个体形态（如Bomhardt, 2017）、躯干形态（如Brinkmann等人, 2015），甚至头部配饰（如Wersenyi, 2005; Wersenyi, 2017）等特定因素，均会对HRTFs产生影响，进而可能影响定位表现与其他听觉感知属性。简言之，头部配饰（如棒球帽、自行车头盔或如今虚拟现实（VR）应用中常用的头戴式显示器）会改变空间听觉线索。 然而在诸多现实场景中，佩戴这类配饰时保持良好的定位能力至关重要——例如骑行时识别驶来的车辆。此外，在使用头戴式显示器的混合现实应用中开展心理声学实验时，必须考虑头戴式显示器对HRTFs的影响。HTC Vive头戴式显示器对定位表现的影响已在Ahrens等人(2018)的研究中得到证实。 若要分析头部配饰在不同入射方向下的影响，需在密集球面采样网格上开展HRTFs测量。但目前搭载各类头部配饰的仿真头HRTF测量数据仍较为稀缺；据我们所知，目前仅有一份针对HTC Vice在含64个测点的稀疏网格上完成的测量数据集可公开获取（Ahrens, 2018）。 本研究提供了两组仿真头的高密度HRTF测量数据：分别为Neumann KU100与HEAD acoustics HMS II.3仿真头，二者均分别搭载自行车头盔、棒球帽、Oculus Rift头戴式显示器，或一副耳外式AKG K1000耳机。本次测量采用VariSphear测量系统（Bernschütz, 2010），可将仿真头精准定位至空间采样点位。所有HRTF数据集均在包含2702个测点的完整球面Lebedev网格上采集。 本研究从频谱特征、双耳线索，以及基于定位模型的定位表现三个维度对实测数据集展开分析，并将结果与未佩戴头部配饰的仿真头参考测量数据进行对比。结果显示，与未佩戴配饰的参考组相比，差异程度因头部配饰类型的不同存在显著差异。针对ITDs与ILDs的分析表明，AKG K1000耳机对其影响最大。对于Oculus Rift头戴式显示器，ITDs与ILDs主要受正面入射声音的影响；而自行车头盔与棒球帽对ITDs与ILDs的影响则极为微弱。 相较于参考组的频谱差异，AKG K1000耳机的频谱偏差最大，Oculus Rift与棒球帽的频谱偏差最小。此外，本研究还分析了受头部配饰影响最显著的声音入射方向：对于Oculus Rift与棒球帽，偏差最大的情况出现在对侧声音入射场景；对于自行车头盔，受影响最显著的方向同样为对侧，但在仰角上有所上移。最终，AKG K1000耳机对实测HRTFs的整体影响最为显著，该影响在声音从后方入射时达到峰值。 本研究的结果适用于需佩戴头部配饰且需考量定位或其他空间听觉属性的应用场景，例如：在混合现实应用中，当听众佩戴头戴式显示器时呈现自然声源；或在特定场景（如佩戴头部配饰的体育运动中）开展定位表现研究。此外，本研究的一项核心目标是提供一套可公开获取的HRTF数据集，该数据集适配听觉可视化（auralization）需求，可用于进一步探究头部配饰对听觉感知的影响。本数据集的HRTF将以SOFA格式（SOFA）公开发布，采用Creative Commons CC BY-SA 4.0许可协议。 ________________________________________________________________________________________________________ 联系方式：Christoph Pörschmann TH Köln - 科隆应用科技大学 通信工程学院 声学与音频信号处理系 Betzdorfer Str. 2, D-50679 科隆，德国 https://www.th-koeln.de/akustik